Download ethics and climate change cost-benefit analysis: stern and after

yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

2009 United Nations Climate Change Conference wikipedia , lookup

Michael E. Mann wikipedia , lookup

Soon and Baliunas controversy wikipedia , lookup

Fred Singer wikipedia , lookup

Global warming wikipedia , lookup

General circulation model wikipedia , lookup

Climatic Research Unit email controversy wikipedia , lookup

ExxonMobil climate change controversy wikipedia , lookup

Heaven and Earth (book) wikipedia , lookup

Climate resilience wikipedia , lookup

Climate change feedback wikipedia , lookup

Effects of global warming on human health wikipedia , lookup

Climate change denial wikipedia , lookup

Climate engineering wikipedia , lookup

Climatic Research Unit documents wikipedia , lookup

Solar radiation management wikipedia , lookup

Economics of climate change mitigation wikipedia , lookup

Attribution of recent climate change wikipedia , lookup

Climate sensitivity wikipedia , lookup

Politics of global warming wikipedia , lookup

United Nations Framework Convention on Climate Change wikipedia , lookup

Climate change adaptation wikipedia , lookup

Effects of global warming wikipedia , lookup

Climate change and agriculture wikipedia , lookup

Climate governance wikipedia , lookup

Climate change in Tuvalu wikipedia , lookup

Climate change in the United States wikipedia , lookup

Carbon Pollution Reduction Scheme wikipedia , lookup

Scientific opinion on climate change wikipedia , lookup

Media coverage of global warming wikipedia , lookup

Citizens' Climate Lobby wikipedia , lookup

Economics of global warming wikipedia , lookup

Public opinion on global warming wikipedia , lookup

Effects of global warming on humans wikipedia , lookup

Surveys of scientists' views on climate change wikipedia , lookup

Climate change, industry and society wikipedia , lookup

IPCC Fourth Assessment Report wikipedia , lookup

Climate change and poverty wikipedia , lookup

Stern Review wikipedia , lookup

Environmental Economy
and Policy Research
Discussion Paper Series
Ethics and Climate Change Cost-Benefit
Analysis: Stern and after
Jonathan Aldred
Number: 44.2009
Ethics and Climate Change Cost-Benefit Analysis: Stern and after
Jonathan Aldred
Abstract: The Stern Review on the economics of climate change (hereafter
‘Stern’) has received much attention regarding its potential political impact. 1 It
has also been extensively discussed among academic economists because
its conclusions are more radical, in terms of action to mitigate climate change,
than previous orthodox economic analyses. This paper aims to contribute to
the debate by critically assessing three key features of climate change CBA,
with particular reference to Stern: monetary valuation of human life,
discounting, and the treatment of risk and uncertainty. This assessment
reveals some common themes, which are examined in the final section. They
concern the tensions which emerge in climate change CBA between ethical
arguments, market-based preferences, and the claims of CBA to scientific
Keywords: Climate change, Stern report, Cost benefit analysis, discounting,
Address for correspondence
Dr. Jonathan Aldred
Department of Land Economy
University of Cambridge
Tel: +44 (0)1223 362329
e-mail: [email protected]
1 Nicholas Stern, The Economics of Climate Change: the Stern Review (Cambridge University Press,
Ethics and Climate Change Cost-Benefit Analysis:
Stern and after
Jonathan Aldred
1. Introduction
The Stern Review on the economics of climate change (hereafter ‘Stern’) has
received much attention regarding its potential political impact.2 It has also been
extensively discussed among academic economists because its conclusions are more
radical, in terms of action to mitigate climate change, than previous orthodox
economic analyses.
But a perhaps more lasting intellectual impact will follow from the explicit
recognition Stern gives to the role of ethical assumptions in cost-benefit analysis
(CBA). On its own, the recognition is not large, and ethical assumptions permeate
Stern much more widely than it acknowledges, but nevertheless it helps to break a
powerful taboo – economic policy analysis should not mention ethics. Before Stern, it
was almost unknown for orthodox economists to debate different ethical assumptions
underpinning economic analysis in general and CBA in particular. There are signs
that this may be beginning to change. 3
This paper aims to contribute to the debate by critically assessing three key
features of climate change CBA, with particular reference to Stern: monetary
2 Nicholas Stern, The Economics of Climate Change: the Stern Review (Cambridge University Press,
3 See for instance Wilfred Beckerman and Cameron Hepburn 'Ethics of the discount rate in the Stern
Review on the Economics of Climate Change', World Economics, Vol.8, No. 1 (2007), pp. 187-210.
valuation of human life, discounting, and the treatment of risk and uncertainty. This
assessment reveals some common themes, which are examined in the final section.
They concern the tensions which emerge in climate change CBA between ethical
arguments, market-based preferences, and the claims of CBA to scientific objectivity.
2. Valuing life
CBA does not value life directly, but in terms of risk of death. The standard
view of market economics is, unsurprisingly, that the money values of life, or
decreased risk of death, should reflect values in markets. On this view, how much
people care about an increased or decreased risk of death can be deduced by
observing their behaviour in markets. In the wage differentials method, economists
observe how much extra people must be paid to persuade them to take a dangerous
job, involving some well-defined risk. If, on average, each person must be paid an
extra £2000 to compensate them for undertaking a job or task involving a risk of
death of 1 in 1000, then economists say that the value of statistical life is £2 million –
£2000 multiplied by 1000. (In the UK, CBAs typically assume a valuation of life of
around £1.5 million, based on wage differentials). 4 An alternative method is simply to
ask people, in surveys, how much they would be willing to pay to reduce some
specified risk, or how much they would require in compensation to tolerate some new
risk. For example, people might be asked how much they would be willing to pay in
increased tax to pay for reducing some hazardous pollutant in city air. Or how much
4 More precisely, £1.145million at 2000 prices. The figure apparently includes gross lost output,
medical and ambulance costs. See The Green Book: Appraisal and Evaluation in Central Government
(HM Treasury, 2003), Annex 2, p. 63.
money would be the minimum acceptable compensation for tolerating the siting of a
hazardous waste dump nearby.
There has been much argument at the IPCC about the monetary valuation of
human life. However, the argument was not about the notion of a money value for
human life per se, but whether lives in rich countries should be assigned a higher
value than those in poor countries. If the raw values from most valuation methods
(such as the survey and wage differentials methods just described) are fed into the
CBA as they stand, then the lives of poor people will be given a lower value than
those of the rich. More specifically, the decreased risk of death, arising from policies
which reduce climate change, attracts a lower monetary value in poor countries.
Compared to rich countries, the extra wage earnt in dangerous jobs will be lower, and
so too will the amount people are willing to pay to avoid an increased risk of death.
On any unadjusted measure of their value drawn directly from market prices or
surveys, poor people will be assigned a lower value. In consultation prior to the
second IPCC report, most governments completely rejected this view, but a group of
economists insisted that the report should include it: ‘A careful reading of the fine
print revealed that they were valuing lives in rich countries at $1,500,000, in middleincome countries at $300,000, and in lowest-income countries at $100,000.’ 5 The
final report heavily qualified this approach, but not because any consensus was
reached. One participant commented: ‘The outcome of it all was that the IPCC is very
reluctant to engage in that controversy again because the proponents on both sides are
still there and obviously still willing to have another fight if the opportunity was given
5 Using 1995 prices. Frank Ackerman and Lisa Heinzerling, Priceless (New Press, 2004), p.73.
to them.’ 6 In subsequent reports, the IPCC has adopted a uniform monetary value of
life of $1,000,000 to apply in all countries. 7
Before turning to whether unadjusted values of life taken directly from market
data should be used in CBA, it is important to see that the issue is broader than the
variations in the monetary value of life which arise due to income differences.
Context-specific valuations of life arise not just because incomes vary, but because the
same ‘objective’ risk – the same probability of harm – is perceived differently in
different contexts. (Whether all situations of uncertainty can be treated as cases of
risk, involving known probabilities, is a question addressed in a later section). In
CBA, the extent of risk is defined solely in terms of its probability, so that two risks
with equal probabilities are regarded as ‘objectively identical’ or ‘statistically
equivalent’. The probability of harm from having an X-ray in hospital is much higher
than that from living near a nuclear power station. But many people regard nuclear
power as riskier than hospital X-rays, because they understand risk in a much broader
sense. Risk has qualities as well as quantities. A risk is perceived to be much worse if
it is involuntary, unfamiliar, unfair, irreversible and uncontrollable; the risks from
nuclear power are widely held to embody all these worrying qualities. These concerns
are now widely discussed in the literature on the social perception of risk; 8 to the list
could be added considerations surrounding the cause of the risk, such as whether it
was due to incompetence, negligence or malice – or natural causes. The breadth,
complexity and interrelatedness of all these factors affecting risk perception, and their
generally qualitative nature, means that there is little hope of capturing them in some
6 Dr. Terry Barker, in oral evidence to the House of Lords Economic Affairs Committee on The
Economics of Climate Change, 22 February 2005.
7 The issue is discussed in IPCC 2001. See B. Metz, O. Davidson, et al., (eds.), Climate Change 2001:
Mitigation (Cambridge University Press, 2001), p. 483. The $1 million value of life is in 1999 prices.
8 See N. Pidgeon, R. Kasperson, and P. Slovic (eds.), The Social Amplification of Risk, (Cambridge
University Press, 2003).
more sophisticated model of quantitative risk assessment. Instead cost-benefit analysts
respond by making ad hoc adjustments to the monetary value of the risk; valuations of
life become context-specific. 9 So, whether due to income differences or differing
perceptions of risk, the same objective risk will be valued differently by different
people, yielding differing, context-specific monetary values of life under any marketbased valuation method.
When conducting climate change CBA, should lives in different countries be
given different, unadjusted, context-specific monetary values, should they be adjusted
in some way, or should a uniform value be used? Contrary to the impression
sometimes given by government manuals on CBA, the ‘measuring rod’ of money is
not neutral. Most obviously, money is more valuable to the poor. More precisely, an
extra unit of money is worth more to the poor than the rich – the marginal value of
money is higher for the poor. This follows directly from the utilitarian ‘law of
diminishing marginal utility’, and has long been recognised in the theory
underpinning CBA. But the value of an additional unit of money will differ across
individuals for other reasons too – ‘materialists’, who especially enjoy the things
money can buy, may value it relatively more highly than ‘ascetics’. 10 Thus monetary
measures of value emerging from market-based valuation methods will not, as they
stand, be comparable across individuals. They should be adjusted to reflect the
differing marginal values of money for different people. Suppose, for example, that
money is three times more valuable to the poor than the rich. Suppose also that in a
poor country, the valuation of life emerging from market-based methods is £500,000;
9 A good example of how ad hoc is given by Jonathan Wolff, 'Risk, Fear, Blame, Shame and the
Regulation of Public Safety', Economics and Philosophy, Vol. 22, (2006), p. 422. The UK Health and
Safety Executive gives all lives a uniform value in its CBA work, except deaths from cancer, which are
valued at exactly twice the usual figure.
10 I am grateful to a referee for reminding me of this point. See K. Brekke, ‘The numeraire matters in
cost-benefit analysis.’ Journal of Public Economics Vol. 64, (1997), pp. 117-23.
but in the rich country it is £2,000,000. To make these numbers comparable (allowing
them to be aggregated in a CBA), the £500,000 figure should be multiplied by three, a
‘distributional weighting’. The adjusted value of life in the poor country is now
£1,500,000: we still do not have a uniform value of life, but the range is narrower.
Although appealing to many academic economists, this approach, involving
adjusted context-specific valuations, is almost never attempted in practice. There are
both theoretical and practical problems involved in obtaining appropriate adjustment
weightings. 11 Trying to determine how much more valuable money is to ‘the poor’
than ‘the rich’ is hard enough, but this is just an approximation; a comprehensive
application of this approach requires that individualized adjustment weightings be
determined for each person, reflecting all the factors which lead the marginal value of
money to vary across people.
There are also practical difficulties in obtaining the (unadjusted) contextspecific valuations in the first place. The wage differentials approach is ruled out.
Wage differentials are context-specific – but the context is different to the one under
consideration. They concern the risk faced by a particular group of workers in a
particular job, not the risk being evaluated in the CBA. This leaves the survey
method, rendered more difficult and costly to conduct, because a fresh survey is
required for each CBA, rather than extrapolating the results from a survey in another
context, the so-called ‘benefit transfer’ method. As well as greatly reducing the cost
of conducting a CBA, practitioners often rely on ‘benefit transfer’ to assist
plausibility: they use valuations obtained in a more plausible context (say, willingness
11 For up-to-date analysis of the theoretical issues, see Olof Johansson-Stenman, ‘Distributional
weights in cost-benefit analysis - should we forget about them?’, Land Economics Vol. 81, No. 3
(2005), pp. 337-52.
to pay higher air fares in return for more stringent aircraft safety requirements), rather
than attempting to ask directly how much people would pay for policy measures
likely to reduce the impact of climate change. Although these concerns are pragmatic,
they are deeply entrenched in CBA. Sugden regards the assumption of ‘contextindependent preferences’ as playing an equally important role as standard
assumptions about the internal consistency of preferences, in making it ‘possible to
gather data in one environment and to draw conclusions about economic value in
another’. 12 More forcefully:
The assumed transferability [of valuations] across contexts … is in fact an essential
condition without which cost-benefit analyses as we know them today cannot be
justified. If all values used in a CBA had to be derived from the precise context of the
particular CBA, then the practice of performing CBAs would come close to that of
performing opinion polls on the topic to be analysed. 13
In short, the goal of fully adjusted, truly context-specific valuations is
probably unobtainable in practice. Instead, many practitioners adopt a uniform value
of life on pragmatic grounds. The idea is that a uniform value will roughly
approximate the adjusted context-specific valuations: in the example above, the
adjusted valuations of life are more alike than the unadjusted ones, the adjustments
acting to ‘cancel out’ the differences in the value of life obtained from market data.
Adopting a uniform value of life assumes, as a short-cut, that this cancelling out is
12 Robert Sugden, ‘Anomalies and stated preference techniques’, Environmental and Resource
Economics Vol. 32, (2005), p.3.
13 Sven Hansson, 'Philosophical problems in cost-benefit analysis', Economics and Philosophy, Vol.
23, No. 2 (2007), p. 179.
Many cost-benefit analysts object to any move away from unadjusted marketbased valuations, whether to adjusted valuations, or the short-cut approach of a
uniform valuation. They argue that determining policy on this basis may lead to a
welfare loss. Suppose again that the unadjusted value of life in poor countries is
£500,000, but that economists input the adjusted valuation of £1,500,000 into a CBA;
the CBA then recommends immediate costly action to address climate change. If
governments in poor countries heed the recommendation, then they may spend more
on tackling climate change than their people think it is worth, as revealed by market
data. That is, the poor country governments would be adopting policies which may
fail an unadjusted cost-benefit test when lives are valued at £500,000. An immediate
response to this argument is that the problem can be solved by ensuring poor country
governments do not bear the (full) costs of their actions to tackle climate change – for
instance, they could receive a subsidy from rich countries. According to this response,
the adjusted CBA still identifies the correct policy choice at a global level; the
problem is simply about who pays. However, defenders of unadjusted market-based
valuations go further. There are likely to be many other policies which would benefit
poor countries more than tackling climate change, policies which pass a cost-benefit
test in the poor country even when lives are valued at £500,000. The subsidy would
be much better spent on supporting (more of) these policies. Subsidising action to
tackle climate change distorts the priorities of poor country governments away from
other actions which their people value more highly.
These arguments rely on debateable assumptions about how welfare changes
are best measured, and whether the status quo distribution of global income should be
taken as given. However, they raise an important question: is the global perspective of
much climate change CBA meaningful, in a world without global government, and
where transfers from rich countries to poor ones are negligible? These are complex
issues, but they do not negate the original objection to unadjusted valuations – money
is not a neutral measure of value. Therefore aggregate unadjusted valuations do not
reflect how we truly value lives: if lives are held to have equal value in different
countries this cannot be captured in a CBA by using different unadjusted monetary
valuations of life across countries. Nevertheless, once we move away from unadjusted
valuations, one of the rationales for putting monetary values on life in the first place
appears to have been lost: CBA may no longer track the policy priorities of people, as
expressed through market data. Abandoning unadjusted valuations may lead us to
abandon the practice of valuing life in monetary terms altogether.
Summing up, it seems that we can reject unadjusted valuations. But adjusted
fully context-specific valuations are unworkable in practice and the global perspective
implicit in making the adjustments may be irrelevant to policy-making. The short-cut
version – a uniform value of life – avoids the practical problems but confronts the
question of relevance even more starkly. A uniform value of life such as the IPCC’s
$1,000,0000 seems abstract and arbitrary, disconnected from the preference data
which is supposed to justify it. At the very least, it seems relevant only once some
prior questions about global justice and responsibility have been settled. The debate
over whether to use context-specific valuations, with or without adjustment
weightings, remains far from settled. 14
14 For further discussion see Cass Sunstein, ‘Are Poor People Worth Less Than Rich People?
Disaggregating the Value of Statistical Lives’, AEI-Brookings Joint Center for Regulatory Studies
Working Paper No. 04-05 (2004). Available at: ; David Anthoff and
Richard Tol, ‘On International Equity Weights and National Decision Making on Climate Change’
FEEM Working Paper, Milan (2007). Available at:
There is yet another way in which context-specific valuations arise, one which
reveals a deeper problem with the very idea of the ‘value of statistical life’. Suppose
that Ann values a reduction in some particular, well-defined risk of her death of 1 in
1000 at £2000. That is, Ann requires £2000 compensation to undertake the risk. Does
this imply that Ann would accept £200000 in return for facing the same type of risk,
but now with a 1 in 10 probability of death? Of course not. The relationship between
probability and value is not linear. Even if we consider the same type of risk, faced
by the same person, it does not follow that valuations are generalisable to new
probabilistic contexts in the way that the CBA literature usually assumes in invoking
a linearity assumption. It is particularly clear that linearity fails in the extreme case:
the amount of money that someone would accept in order to face certain death cannot
be calculated by scaling up valuations for very small risks of death. The idea is
absurd, and seems to have no relevance to the practice of CBA: defenders of CBA
emphasise that it is concerned with ‘statistical’ lives lost, not real ones. But policies
evaluated using CBA often involve real lives being lost, predictably so. Even if the
policy imposes only a small increased risk of death on each individual, real lives are
highly likely to be lost whenever very large numbers of people are affected, or the
risks are repeatedly imposed over a long period. In these cases, by the ‘law of large
numbers’ from probability theory, we can be confident about roughly how many will
die. Climate change provides many examples – relatively small increases in the risk of
death faced by millions of people.
No argument is being made here against policies which lead to lives being
lost. Sometimes such policy choices are correct, because the alternatives are worse.
The point is that the CBA treatment of these policies is exactly equivalent to putting a
money value on real lives directly, without the ‘statistical’ smokescreen. For example,
suppose that due to climate change, 60 million people (the UK population) each face a
one in one million increase in the risk of death. Each person’s increased risk is valued
at £2 by the CBA, so the total risk would be valued at £120million. We can
confidently predict in advance that on average 60 people will die. So it seems more
transparent, and less misleading, to state simply: on average 60 people will die, and
the CBA values their lives at £2million each.
The argument here is about more than terminology. First, even supporters of
CBA are concerned about these kinds of cases. Whether obtained through surveys or
revealed in the market through methods such as wage differentials analysis, individual
valuations for risks as low as one in one million are likely to be unreliable, due to well
known cognitive biases in interpreting small probabilities. Valuations would probably
be very low, sometimes zero, implying that CBA might assign a very low total value
to the risk, allowing the policy to pass the cost-benefit test, even though on average 60
people will die. This is the average impact; depending on how the risk manifests
itself, there could be a catastrophe in which many more would die. Although no worse
for each of those who die, the overall impact on society of this catastrophe would
probably be disproportionately greater. 15
Second, valuations of small risks should not be used to inform climate change
policy when real lives are predictably lost, because linearity is false. The problems
here become particularly stark when a third issue, identity, is raised. Consider a
hypothetical policy which would lead to the death of an identifiable person, John.
Clearly the money value which John would place on his own life would be infinite,
15 Richard Posner, Catastrophe (Oxford University Press, 2004), p. 165-70; Cass Sunstein, Laws of
Fear (Cambridge University Press, 2005), p. 160-1. Neither Posner nor Sunstein seem to realise that
this example reveals a general flaw with valuing life in terms of risk, not just one applying to small
probabilities of catastrophe.
consistent with the theory behind CBA. There is no amount of money which John
would be willing to accept as compensation for his death. So because of John’s death
alone, the costs of this policy are infinite, and it would fail the cost-benefit test.
Contrast this case with the policy just described, which might pass a cost-benefit test,
even though 60 people die on average. This is possible only because the identity of
those who will die is unknown, so there is no particular individual like John. It is
absurd that ignorance should make a difference here. What matters is the numbers
who will die, not whether we know their identity. The ‘value of a statistical life’ is a
sleight-of-hand device to conceal this absurdity. Presenting the policy as ‘causing the
death of 60 people on average, but we do not know who’ may make it much harder
for cost-benefit analysts to persuade their audience that a money value could be put on
the risk. Instead, the risk is represented as a loss of statistical lives, or increased
probability of death. But it is still the same loss, and we should not be deceived into
believing it can be valued in terms of money. 16
3. Discounting
Discounting concerns the idea that future costs and benefits should have less
weight in our decision-making than present ones. It is difficult to overstate the
importance of discounting in the economic analysis of climate change. Almost all of
the debate among orthodox economists following the Stern Review has concerned its
treatment of discounting. It is easy to see why: the results of climate change CBA are
acutely sensitive to the discount rate chosen. Stern’s effective discount rate is 1.4 per
cent, while many economist critics have advocated a rate around 6 per cent. ‘The
16 The identity argument first appeared in John Broome, ‘Trying to value a life’, Journal of Public
Economics Vol. 9 (1977), pp. 91-100.
present discounted value of a given global-warming loss from a century hence … [at 6
per cent] is one hundredth of the present discounted value of the same loss at Stern’s
annual interest rate of 1.4%.’ 17 And yet, for many non-economists, discounting seems
obviously inapplicable to climate change policy, because it has an extremely
implausible implication: huge impacts in the distant future count for almost nothing in
our current decision-making. ‘Suppose that Denmark needs to be evacuated due to
flooding. Current real estate value in Denmark is estimated at about $238 billion. If a
discount rate of 5% is applied, then over 500 years, the same real estate would be
worth just $6.’ 18 And here is another compounding effect: since human lives are
valued in CBA, and CBA discounts all impacts, then future lives are discounted.
Discounting at six per cent, one life now is worth 339 lives in 100 years time.
If these examples lead us to adopt a uniform discount rate of zero instead, not
only is discounting redundant, but formal climate change CBA too, because the result
of a CBA is apparent without any formal analysis. Simply because the costs of
adverse climate change are felt by so many generations, extending into the far future,
they will substantially outweigh the costs of mitigation borne now. Most economists
go further: no discounting implies we should reduce our income to a subsistence level
now for the sake of future generations. Any sacrifice now will be outweighed by the
benefit to future generations, because investments made now will grow and yield
greater benefit in the future, and there are many future generations who will benefit.
Economists regard this excessive sacrifice argument as a reductio ad absurdum
against zero discounting, although there are convincing reasons to doubt the excessive
17 Martin Weitzman, 'A Review of The Stern Review on the Economics of Climate Change', Journal of
Economic Literature, Vol. 45, No. 3 (2007), p. 708.
18 Stephen Gardiner, 'Ethics and global climate change', Ethics, Vol. 114 (2004), p. 572.
sacrifice argument. 19 But if CBA applied in cases involving large very long-term
impacts generates absurd recommendations both with zero discount rates and positive
ones, then it is tempting to conclude that discounting methods are irrevocably flawed,
at least in these cases, and should be abandoned. However, since economists are only
convinced of the absurdity in the case of zero discounting, they continue with the
practice. Against this background, and given what is at stake, it is worth exploring the
core arguments about discounting in some detail.
The current debate over discounting largely concerns two arguments for it:
1. Pure time preference. This is the claim that, all other things being equal,
people prefer welfare or satisfaction now to the same amount of it later. The claim is
usually justified by reference to individual impatience.
2. Future consumption is worth less. Economic growth means that
consumption levels will be higher in the future, so by the law of diminishing marginal
utility, the satisfaction from additional future consumption will be lower.
In comparison to orthodox economic theory or past practice in CBA, Stern
significantly limited the scope and magnitude of discounting, and attracted substantial
flak from many economists for doing so. This occurred despite Stern devoting much
more attention to economists’ arguments that it discounts too little, than the opposite
view of philosophers and others, that it still discounts too much. 20 On ethical grounds,
19 For critique of the excessive sacrifice argument see John Broome, Counting the Cost of Global
Warming (White Horse Press, 1992); Tyler Cowan and Derek Parfit, ‘Against the Social Discount
Rate’, in Peter Laslett and James Fishkin (eds) Philsophy, Politics and Society: Future Generations
(Yale University Press, 1992).
20 See especially Stern, The Economics of Climate Change, pp. 35-7, 50-4, 59-60.
Stern rightly rejects the argument from pure time preference altogether. 21 That is,
Stern is impartial between generations in the sense of attaching equal weight to
welfare impacts, whether they are experienced now or by future generations. Stern
defends the second argument, although emphasises that it does not apply uniformly
across goods and people. It is likely that some future people will be worse off,
because increased inequality will outweigh overall economic growth. Also, while
some goods will be more abundant in the future, others will be equally scarce or more
so, because economic growth cannot generate more of them. Stocks of non-renewable
resources such as fossil fuels will decline; the same is arguably true of unique,
irreplaceable environmental assets such as beautiful countryside. The upshot is that
the discount rate for these goods, or for impacts affecting groups who are worse off in
the future, should be zero or negative as appropriate.
Stern also responds to a particular objection to discounting when future
generations are affected. It is not a new objection, but until climate change issues
became prominent, most cost-benefit analyses assumed that all the major impacts of
CBA-based decisions are felt within thirty years or so – that is, by the current
generation. 22 Climate change means the objection can no longer be ignored. The
problem is that future generations are disenfranchised. 23 In CBA, the value assigned
to the discount rate is influenced by market interest rates. (The influence varies
according to differing views about how discount rates should be derived, as discussed
21 Stern rejects the argument for pure time preference based on impatience, but argues that there is a
non-trivial probability that the human race may not recognisably survive into the distant future, because
of catastrophes like a meteorite impact or devastating nuclear war. This leads Stern to assume a small
degree of effective pure time preference: pure time preference equivalent to the human race facing an
almost 1 in 10 chance of extinction within 100 years, which seems implausibly large if this is really the
only justification. See Stern, The Economics of Climate Change, pp. 35-6, 53.
22 For instance, guidance for the conduct of CBAs in previous editions of the UK Treasury Green
Book recommended time horizons of around 30 years.
23 This version of the argument that future generations are disenfranchised draws heavily on John
Broome, Ethics out of Economics (Cambridge University Press, 1999), Chapter 4.
in the next section.) At least in principle, market-determined rates reflect preferences
– specifically, how impatient we are. But only the preferences of current generations
are reflected in markets; the preferences of future generations are represented only
insofar as current generations care about them. Future generations are disenfranchised
just in the way that women are disenfranchised in societies where only the views of
men have any impact on the decision. This may seem like the familiar point about
intergenerational justice requiring impartiality between generations. But the present
argument is different: even if such ethical considerations turn out to sanction the
practice of discounting in principle, in practice we should not let current interest rates
influence calculation of the discount rate, because only current generations have any
sway over these interest rates. Obviously, it is inevitable that only current generations
make decisions – in that sense future generations are unavoidably disenfranchised –
but it is less obvious, and not inevitable, that this bias is reinforced by relying on
market interest rates. Clearly the argument here goes beyond interest rates. An
identical objection can be made to using observed saving rates to determine the rate of
pure time preference (even if a positive rate of pure time preference is otherwise
regarded as legitimate).
In response to this argument, rather than abandon discounting, economists
have characteristically embraced a technical modification with great enthusiasm. This
new work in economic theory is seen as answering the sceptics, and has swiftly turned
into UK Government policy, now reflected in Stern. 24 The proposal is that discount
rates should decline over time, rather than remain constant as in standard CBA. With
24 The Green Book, 2003. The UK Office of the Deputy Prime Minister commissioned a report on
recent research on discounting, which explicitly recommended declining discount rates as a response to
the philosophical objections: A social time preference rate for use in long-term discounting (Office of
the Deputy Prime Minister, 2002), p. 13. The Stern Review shares this conclusion, although for
different reasons. See Stern, The Economics of Climate Change, pp. 35-7, 56-7.
declining rates, the distant future counts for much more than under conventional
discounting, so that impacts affecting future generations are discounted less. But this
more favourable treatment of future generations is a by-product of declining discount
rates, not their rationale, which concerns uncertainty. The rationale runs as follows.
Until very recently, if they did not know what discount rate to pick, economists
simply adopted an average (possibly weighted) of the plausible rates, constant over
time. It is now known that this strategy reflects a mathematical error. The costs and
benefits are actually multiplied by discount factors, so it is these factors which should
be averaged in response to uncertainty, not the rates themselves; averaged discount
factors imply, counter-intuitively, a much lower effective discount rate, and one
which declines over time. 25
How does this new approach claim to answer the ethical objections to
discounting? Ethical doubts about the discount rate are interpreted as a form of
uncertainty. For example, a ‘materialist’ might favour a high discount rate, while a
‘conservationist’ would favour a low one. 26 Then we should adopt declining discount
rates over time in order to ‘average’ the weight given to the two ethical positions.
While sceptics about discounting may welcome the belated recognition given to
competing ethical views, the overall approach is a bizarre one. Trying to ‘average’ the
implications of ‘materialist’ and ‘conservationist’ views is akin to the person who,
because they are unsure whether to become a Christian or a Buddhist, decides to be a
25 The discount factor in period t, δt, is related to the discount rate r using the standard net present
value formula δt =1/(1+r)t. A pioneering theoretical contribution on declining discount rates is Martin
Weitzman, 'Why the far distant future should be discounted at its lowest possible rate', Journal of
Environmental Economics and Management, Vol. 36 (1998), p. 201-8. For a simple example
illustrating the mathematical relationships which give rise to declining discount rates, see D. Pearce, B.
Groom, C . Hepburn and P. Koundouri, 'Valuing the Future', World Economics, Vol. 4, No. 2 (2003),
pp. 128-9.
26 C. Li and K. Lofgren, ‘Renewable resources and economic sustainability’, Journal of
Environmental Economics and Management, Vol. 40 (2000), pp. 236-50.
Buddhist on Mondays, Wednesdays and Fridays, and a Christian on Tuesdays,
Thursdays and Sundays.
To sum up, CBA aims to provide guidance on what we should do, so our
obligations to future generations matter. Discounting disenfranchises future
generations, and the technical fix of declining discount rates does not address the
problem. Debates over discounting bring ethical issues to the fore, but discounting
may not offer the appropriate framework for the resolution of difficult ethical
dilemmas raised by problems of intergenerational justice. At best, discount rates
express the resolution in simplified form; they do not help to reach it.
4. Risk and uncertainty
It is worth stressing the challenge which uncertainty poses to climate change
policy-making, because some suggest that the uncertainty is no greater than in many
other policy problems. Climate change uncertainty is driven by distinctive features.
There are complex feedback effects which cut back and forth between the social and
natural worlds. For example the scale and rate of climate change will affect economic
growth and the speed of technological innovation, both of which in turn influence
emissions levels, feeding back to shape climate change. Uncertainty about the causes
of climate change is not just ex ante but ex post. That is, there is no event which
resolves the uncertainty. There may be many more hurricanes than previously, but
there is no way of knowing whether any particular hurricane is due to climate change.
Uncertainty is only partly a result of incomplete knowledge or lack of research.
Rather, research has shown that uncertainty is intrinsic to the problem. One reason is
that increased uncertainty about the climate is itself a consequence of climate change,
not just an exogenous constraint. Another is that climate change manifests itself in a
highly non-linear, chaotic system which is inherently unpredictable. For example, a
characteristic of chaotic models is their extreme sensitivity to parameter values –
popularly illustrated by the ‘butterfly effect’. In sum, it is hard to see what other
policy issue shares all these features of uncertainty. 27
Stern represents an improvement on past practice in CBA in terms of its
treatment of uncertainty. The Keynesian/Knightian distinction between risk and
uncertainty is explicitly acknowledged: under risk, the full range of possible outcomes
is known, along with their probabilities; under uncertainty, either the range is
unknown, or the probabilities, or both. (This meaning of uncertainty will be termed
pure uncertainty hereafter). As already indicated, in climate change analysis, pure
uncertainty does not arise only when there is an absolute blank in our knowledge, but
as an intrinsic phenomenon, or because scientists fundamentally disagree about
probabilities or impacts.
Many orthodox economists endorse some form of Bayesianism. That is, there
is no such thing as pure uncertainty in the sense I have defined it. Decision-makers
always use probabilities, consciously or otherwise, when outcomes are not certain.
Most orthodox economists who reject these claims nevertheless adhere to a weaker,
normative version of Bayesianism: rational choice requires that we must attach
probabilities, explicitly or implicitly – so it is better to be explicit about it. Although
distinct, Stern’s ‘subjective preference’ approach remains close to this view, despite
27 See Lisa Heinzerling, and Frank Ackerman, 'Law and Economics for a warming world', Harvard
Law and Policy Review, Vol. 1, No. 2, (2007), pp. 331-62.
its acceptance of the possibility of pure uncertainty. 28 Both of the techniques which
Stern uses in response to this possibility involve quantifying the uncertainties.
First, Stern uses an econometric technique (‘Monte Carlo simulation’) to show
how the results of his core model are highly uncertain: they vary dramatically
depending on the assumed parameter values. However, for each of the 31 parameters
in the model, Stern still assumes we know both the exact range of possible values, and
the probability distribution of the parameter. 29 Second, in cases of pure uncertainty,
Stern does not demand probabilities for all the unknowns, but does assume we know
the exact range of possible probabilities, and can put numerical weights on these
possible probabilities according to factors such as ‘the magnitude of associated
threats, or pessimism, and possibly any hunch about which probability might be more
or less plausible.’ 30 The result of these assumptions is a kind of quantitative pseudoprobability attached to each uncertainty. Stern’s approach here suffers from the same
defects as Bayesianism applied to CBA. The numerical weights to be attached to
probabilities must be invented, forcing inherently qualitative views about catastrophic
risks and the like into an unhelpful quantitative framework. With both techniques, the
outcome of the CBA will have a misleadingly reassuring precision, implying that we
know more than we do.
Stern also claims that the numerical weights which form part of the
quantitative pseudo-probabilities enable the precautionary principle to be incorporated
into CBA. The idea appears to be that if some possible outcome is particularly
catastrophic, then a greater weight can be attached to the probabilities associated with
28 Stern, The Economics of Climate Change, pp. 37-9.
29 Ibid., p. 173.
30 Ibid., p. 39.
it. This attempt to give catastrophe risks special treatment in the analysis is welcome,
and it is accompanied by various ‘special mentions’ throughout the Review.
Nevertheless, it is unconvincing from several perspectives. To begin with, Stern’s
approach advises the decision-maker to ‘act as if she chooses the action that
maximises a weighted average of the worst expected utility and the best expected
utility.’ 31 This does not capture the underlying intuition behind the precautionary
principle, that when the worst outcome could be severe, we should take all reasonable
steps to avoid it. No version of the precautionary principle involves trading off the
best and worst outcomes against each other. We return to the precautionary principle
Support for the view that Stern understates the severity of uncertainty also
comes from a very different source. Harvard economist Martin Weitzman is a leading
figure in orthodox economics who has made innovative theoretical contributions, such
as his work on declining discount rates outlined above. Weitzman begins by noting
that IPCC 2007 considers six ‘equally sound’ scenarios for global average
temperature increase. There is pure uncertainty, so
IPCC does not attach
probabilities, but if these ‘equally sound’ scenarios were ‘equally likely’ then it
follows that there is a probability of around 3 per cent of global warming of 6ºC or
more within 100 years. 32 Since uncertainty means such non-trivial probabilities
cannot be ruled out, and given the horrific effects of 6ºC warming, Weitzman argues
that catastrophes should be central to any analysis. Consistent with his view that CBA
parameters should be derived from market data, Weitzman looks to behaviour in
markets for clues on how to treat catastrophe risks. Specifically, he argues that
31 Ibid., p. 39.
32 Weitzman, ‘A Review of The Stern Review’, p. 716.
various puzzling patterns of behaviour in financial markets – so-called ‘asset return
puzzles’ – only make sense if
‘investors are disproportionately afraid of rare
disasters’ 33 ; and he adds a theoretical econometric argument suggesting that this
disproportionate fear is legitimate in a world involving pure uncertainty. Weitzman
concludes that the damages in case of catastrophe will generally be decisive in
determining the recommendation of any kind of CBA, largely regardless of its other
parameters such as the discount rate. Applying this framework to climate change, two
important points emerge. First, even economists persuaded of the wisdom of
attempting global CBA of average impacts of climate change must recognise that
impacts in a world with 6ºC warming ‘are located in the terra incognita of what any
honest economic modeller would have to admit is a planet Earth reconfigured as
science fiction.’ 34 It is simply absurd to attempt to measure these impacts in monetary
terms. Second, no such attempt is necessary, even from a narrow CBA perspective,
because there are good reasons in economic theory to justify very substantial costs to
eliminate the risk of uncertain disasters. Weitzman summarizes these arguments with
the following contrast. Standard climate change CBA is about ‘consumption
smoothing’ – trading-off sacrifices in consumption now for the sake of benefits in the
future. Instead we should think in terms of ‘catastrophe insurance’.
To non-economists, and many economists, Weitzman’s analysis may seem
idiosyncratic. But it may prove crucial in shifting the terms of debate in orthodox
economics. At the very least, it makes one fundamental claim. Economic analysis of
climate change is preoccupied with average outcomes: attempting to assess the
average effects, if the Earth experiences the average (i.e. most probable) amount of
33 Ibid., p. 715.
34 Ibid., p. 716.
climate change. But if averages cannot be calculated, then this analysis is misleading
and futile. Instead we must consider all possible outcomes, rightly and inevitably
paying the most attention to possible catastrophes.
These considerations return us to the precautionary principle. There has been a
voluminous literature on the PP, much of it critical. But some of the harshest critics
have recently been persuaded that a more coherent PP can be recovered, one which
escapes their objections. 35 This ‘core PP’ has a much narrower range of application,
inspired by Rawls’ criteria to determine the applicability of the difference principle. 36
The precautionary principle is analogous to Rawlsian ‘maximin’ because in both
approaches, priority is given to the worst off/case rather than trading off this priority
against others. Under the core PP, precaution is justified when three conditions are
met: (i) some plausible possible outcomes are catastrophic; (ii) probabilities cannot be
attached to these outcomes; (iii) the loss from following maximin is relatively small.
It is clear that climate change meets the first two conditions; regarding the third, much
hangs on the word relatively. Essentially, precaution is justified if the costs of
aggressive action to stabilize carbon emissions are small relative to the size of the
potential catastrophe. Many would argue that this is true, if the costs can be
understood as around 2 per cent of global GDP. But of course such cost estimates are
still vulnerable to the criticisms of climate change CBA made here and elsewhere.
Plausible versions of the precautionary principle do not, on their own, allow us to
escape from standard cost-benefit thinking. On the other hand, if we make the ethical
judgement that the effects of a climate change catastrophe would be qualitatively
35 Cass Sunstein is one prominent critic who now supports a limited precautionary principle. See
Sunstein, Laws of Fear, chapter 5.
36 See especially Stephen Gardiner, 'A Core Precautionary Principle', Journal of Political Philosophy,
Vol. 14, No. 1, (2006), pp. 33-60.
different from, and worse than, the consumption losses involved in taking action now
– immeasurably worse – then aggressive action can be justified. But as the next
section argues, most orthodox economists are deeply reluctant to admit explicit ethical
judgements, despite their ubiquity in CBA. 37
5. Revealed preference, ethics and cost-benefit science
In light of the problems discussed in the previous three sections, CBA appears
to be a seriously flawed tool for guiding decisions about climate change policy. There
is a large and vitally important question about the extent to which governments should
continue to rely on CBA for these purposes. But given its enormity, I will approach
just one aspect of this question indirectly, examining the way in which arguments
about climate change CBA are conducted.
‘Revealed preference’ is the conceptual ghost which haunts all aspects of the
analysis. Individual choices in markets supposedly reveal people’s preferences, which
are then used by economists to deduce important parameter values in their models.
These parameters can have a decisive influence on the recommendations of CBA. In
the above discussion of the Stern Review and related work, there are many examples
of the attempt to derive parameter values from observed market behaviour, including
deriving the discount rate from observed saving rates and interest rates, and
determining monetary valuations, such as the value of life, using revealed preference
methods such as the wage differentials approach. The practice of inferring
37 For further discussion of the reluctance of economists to acknowledge ethics, see Daniel Hausman
and Michael McPherson, Economic Analysis, Moral Philosophy and Public Policy, (Cambridge
University Press, 2006); Philippe Mongin, 'Value Judgments and Value Neutrality in Economics',
Economica, Vol. 73 (2006) pp. 257-86.
probabilities in Bayesian fashion from market choices is another example of revealed
preference reasoning. Although discussion in economics often proceeds as though
there is no alternative to revealed preference approaches, debates about discounting
show more awareness of the issues.
At least since influential work by Arrow and Stiglitz for IPCC1995, there has
been explicit recognition of ‘descriptive’ versus ‘normative’ approaches to
discounting. 38 The descriptive approach seeks to derive discount rates from market
data; the normative one appeals to ethical principles to determine the rate. Arrow
emphasises the merits of the normative approach, but nevertheless takes what might
be termed a ‘weak descriptive’ view: even if market data alone cannot determine the
discount rate, our ethical choices must be ‘consistent’ with observed market
behaviour. Perhaps the most notable feature of the debate following Stern, and
surrounding climate change CBA more generally, is the conflict over whether to take
a descriptive or normative approach to discounting. The debate has been
characterised, by those on both sides, as reflecting a fundamental difference of view
between UK and other economists. 39 The language of the debate is sometimes
intemperate: Nordhaus lambasts Stern for adopting ‘the lofty vantage point of the
world social planner, perhaps stoking the dying embers of the British empire’. 40 On
the UK side, Dasgupta has endorsed Stern’s approach, which in its willingness to
reject market data on ethical grounds, can be traced back to Ramsey, via Sen. 41
38 K. Arrow, W. Cline, K.-G. Maler, M. Munasinghe, and J. Stiglitz, ‘Intertemporal equity,
discounting, and economic efficiency’, in J. Bruce, H. Lee and E. Haites (eds) Climate Change 1995:
economic and social dimensions of climate change, (Cambridge University Press, 1996), p. 125.
39 See for instance Weitzman, ‘A Review of The Stern Review’, p. 703-24; Angus Deaton, 'On
transatlantic vices, or Stern in America', Royal Economic Society Newsletter, Vol. 139 (2007), p. 3-4.
40 William Nordhaus, 'A Review of The Stern Review on the Economics of Climate Change', Journal
of Economic Literature, Vol. 45, No. 3, (2007), p. 691.
41 Partha Dasgupta, 'The Stern Review's Economics of Climate Change', National Institute Economic
Review, Vol. 199, (2007) p. 4-7.
Elsewhere, the majority revealed preference view among economists was adopted by
Lomborg’s ‘Copenhagen Consensus’ panel, and endorsed again in the aftermath of
Stern by Arrow, Nordhaus, and Weitzman, among others. 42
Despite their persisting popularity among economists, revealed preference
arguments are fundamentally flawed; the ‘weak descriptive’ view of Arrow hardly
fares better. To begin with, choice cannot reveal preference because choice depends
on both belief and preference. 43 Only with comprehensive information about beliefs
can we hope to infer a preference from an observed choice. For example, wage
differentials cannot alone form the basis for valuing life. The observing economist
would also need to know how workers think about risks. Specifically, the economists
would need to know, or assume, that workers consciously weigh up risks against
wages, have accurate probability information about the risks they face, construct their
beliefs about risk on the basis of this probability information (rather than ‘trusting to
fate’), and do not suffer from common cognitive biases in their apprehension of
probabilities. Similarly, observed saving rates reflect habit, inertia, trends in house
prices, and beliefs about future inflation and interest rates, as much or more than they
reveal pure time preferences between present and future consumption. These
examples raise a second worry: even if the observing economist somehow knows
enough about the individual’s beliefs to derive her preferences from her choices, the
standard ‘consumer sovereignty’ justification for relying on these preferences breaks
down if the preferences are based on false beliefs, incomplete information or
cognitive errors. That is, policy based on a correct reading of preferences from
42 Kenneth Arrow, ‘Global Climate Change: A Challenge to Policy’, The Economists' Voice, (2007);
Lomborg, Global Crises, Global Solutions; Nordhaus, 'A Review of The Stern Review’.
43 For seminal contributions in economics and philosophy see Amartya Sen, 'Behaviour and the
Concept of Preference', Economica, Vol. 40, (1973), pp. 241-59; Donald Davidson, 'Actions, reasons
and causes', Journal of Philosophy, Vol. 60, (1963), pp. 685-700.
choices may still fail to be welfare-maximising. More generally, recent work in
behavioural economics provides overwhelming evidence that individuals are not good
judges of their own interests, so welfarist appeals to consumer sovereignty are
spurious. A third problem is the context-specificity of preferences. We have seen this
in the case of valuing life but the problem is more general. Preferences revealed
through market data may be entirely different from those the economist is actually
seeking to uncover – the preferences in the non-market context of climate change
Regardless of the status of revealed preference arguments, it does not follow
that the preferences so revealed should guide policy. Individual preferences expressed
in markets reflect the narrower concerns of individuals as consumers, whereas climate
change policy demands the preferences – or rather, judgements – of individuals as
citizens who consider what society should do. For example, while individuals may
exhibit impatience (pure time preference) in their behaviour as consumers in markets,
preferring consumption now to the same amount later, as citizens they may judge pure
time preference to be wholly inappropriate to societal decision-making. So even if we
decide that discount rates should be determined by aggregating the judgements of
members of society, these judgements cannot be derived by reading off preferences
from market behaviour. Climate change policy also raises particular ethical
considerations which take us beyond market data: this data can only reflect the
preferences of current generations, so future generations are disenfranchised. This
point was noted above with regard to using market interest rates and savings rates to
guide the setting of discount rates, but again, the objection to market data is more
In light of all these objections to using market data to determine valuations
and model parameters, it is worth asking why most orthodox economists still do so.
An important part of the explanation appears to be a widespread determination to
deny the role of ethics, in favour of economics as ‘science’. Nordhaus finds ‘ethical
reasoning on discount rates [in Stern] largely irrelevant’. 44 In an acerbic passage,
Weitzman is equally contemptuous, stating that Stern (and Cline, a precursor also
adopting a normative approach) rely:
mostly on a priori philosopher-king ethical judgements about the immorality of
treating future generations differently from the current generation—instead of trying
to back out what possibly more representative members of society than either Cline
or Stern might be revealing from their behavior is their implicit rate of pure time
preference. An enormously important part of the ‘discipline’ of economics is
supposed to be that economists understand the difference between their own personal
preferences for apples over oranges and the preferences of others for apples over
oranges. Inferring society’s revealed preference value of
[rate of pure time
preference] is not an easy task … but at least a good-faith effort at such an inference
might have gone some way towards convincing the public that the economists doing
the studies are not drawing conclusions primarily from imposing their own value
judgements on the rest of the world. 45
These comments are all the more remarkable because Weitzman appears to
take the more nuanced ‘weak descriptive’ approach of Arrow; Weitzman is certainly
aware that actual market behaviour often appears inconsistent with economic theory
(for instance in his ‘asset return puzzles’), so that revealed preference methods cannot
be applied straightforwardly. In short, Weitzman appears much more concerned to
exclude ethics than to insist on using market data alone to derive discount rates.
44 Nordhaus, 'A Review of The Stern Review’, p. 692.
45 Weitzman, ‘A Review of The Stern Review’, p. 712.
Among economists who favour a normative approach, there is still unease about
acknowledging the role of ethics. Although Dasgupta has written on some
philosophical issues around discounting, he appears to regard such discussion as
inappropriate within economics: he has devoted an entire paper to arguing that
economists argue about facts, not values. 46
More generally, both Nordhaus and Weitzman dismiss the Stern Review as
being much more a ‘political document’ than a scientific analysis, making repeated
references to economics being a science. 47 We have seen the scientific pretensions of
climate change CBA elsewhere, in the focus on quantification and consistency.
Quantification goes far beyond the realm of the market, making progressively more
heroic assumptions to extract numbers, as in the treatment of pure uncertainty and the
monetary valuation of life. The practice of valuing life is also dominated by the drive
for consistency, as is the practice of replacing subjective judgements about risk with
objective probabilities whenever those are available.
At this point, there is no need to confront difficult questions about the nature
of science, because the project to present CBA as an ethics-free science is internally
flawed. Firstly, CBA clearly rests on a comprehensive ethical framework, perhaps
best labelled ‘welfarism’. This may appear obvious, but it is notable that Nordhaus, in
appealing to alternative ethical systems to support his belief in pure time preference,
does not acknowledge that such systems would typically abandon CBA altogether as a
46 Partha Dasgupta, ‘Three conceptions of intergenerational justice’, in H. Lillehammer and D.H.
Mellor (eds), Ramsey’s Legacy (Oxford University Press, 2005); Partha Dasgupta, 'What do
economists analyze and why: values or facts?' Economics and Philosophy, Vol. 21, No. 2 (2005), pp.
47 See for instance Nordhaus, 'A Review of The Stern Review’, p. 688; Weitzman, ‘A Review of The
Stern Review’, p. 723.
guide to social decision-making. 48 It is as though CBA per se (as opposed to
normative approaches to discounting) can be conceived in ethics-free terms, without
appeal to its welfarist foundations. Recall that many economists still believe CBA can
be justified in terms of Pareto efficiency alone, and ‘efficiency’ is sometimes seen as
a matter of fact, not value. 49
This raises a second, more fundamental difficulty. There is little recognition in
orthodox economics that even if first order ethical judgements are sidestepped, second
order judgements are made about the institutional setting for those first order
judgements. Effectively, a kind of meta-ethical judgement replaces the first order
ethical judgement. For example, the decision to determine discount rates and other
parameters on the basis of market data is in this sense a meta-ethical judgement to
place the ethical choice in the hands of the market. Some economists appear barely to
notice making this judgement because they do not acknowledge there is a choice to be
made: there is no alternative. That is, there is almost no recognition that legitimate
policy-making can proceed on a different basis to private decision-making, that
factors relevant to one may not be relevant to the other, so that public and private
decision-making can be inconsistent in the sense of ‘revealing’ different implicit
parameters (such as different discount rates) without either decision process being
Third, at the level of practice, CBA is unavoidably riddled with ad hoc ethical
assumptions if monetary values are to be assigned. This is less widely acknowledged
than welfarism: recognition of the role of ethics at this level does not usually extend
beyond the choice of distributional weights and the discount rate. It is not just the key
48 Nordhaus, 'A Review of The Stern Review’, p. 692.
49 For careful exposition of the flaws in this view, see Hausman and McPherson, Economic Analysis.
parameters which depend on ethical assumptions, but the minor assumptions too,
which are acknowledged, if at all, in a footnote. Two examples must suffice here.
Stern makes modelling assumptions which rule out ‘consumption externalities’ such
as envy. 50 ‘Envy’ may appear undesirable, but Stern’s assumption rules out the
possibility that the satisfaction from material consumption may depend on the
consumption levels of others, a possibility backed by substantial psychological
evidence. In providing guidance for CBAs of transport projects, the U.K. Treasury
Green Book assumes that time spent sitting in a traffic jam is less unpleasant (i.e. less
costly) than time spent walking or waiting at a bus stop. 51 While these kinds of hidden
assumptions may not have much impact on the overall analysis individually, the
cumulative effect may be significant. This is more likely if the assumptions, although
apparently independent, are informed by a common normative perspective, perhaps
concerning the preferences people ought to have, or the belief that the market is
usually the best institution for resource allocation.
Fourth, another reason for the ubiquity of ethical assumptions is that
economists sometimes reject market data even when it is available and unambiguous.
As already noted, there is a tension between economists’ loyalty to market data –
consumer sovereignty – and the desire for scientific consistency and objectivity. A
related conflict arises between market data and economic theories of rationality,
which have a normative element. If the market data reveals ‘irrational’ preferences,
then some economists side with the raw preferences: pure time preferences and
50 Stern, The Economics of Climate Change, p. 56.
51 The Green Book, 2003, Annex 2.
preferences about subjective risks are treated like any other preferences. 52 Others
favour the theoretical account of rationality, and accordingly abandon, modify or
otherwise launder the data so that it more closely reflects the preferences market
participants would have, if they were ‘fully rational’.
Once the full range of value judgements is acknowledged, then climate change
CBA becomes recognised as a quantitative framework for structuring ethical decisionmaking – applied moral mathematics rather than science. Many facts in CBA are
inextricably entangled with values. These facts can only be stated in terms which are
partly evaluative: description and evaluation cannot be separated. 53
It is worth stressing what is not being argued here. There is no suggestion that
facts and values are indistinguishable. Or even that cost-benefit analysts cannot make
value-free statements (although they may be hard to identify in practice). 54 Although
ethics cannot be stripped out of CBA, it does not follow that CBA is redundant.
Rational argument about ethical judgements – whether within the framework of CBA
or not – is not only possible, but central to policy-making.
6. Concluding remarks
What role, if any, should CBA play in informing climate change policy? One
reply is that applications of moral mathematics such as CBA have a distinctive role to
play in clarifying the nature of ethical disagreements, facilitating the kind of rational
52 An influential defence of this view is John Harsanyi, ‘Morality and the theory of rational
behaviour’, in Amartya Sen and Bernard Williams (eds) Utilitarianism and Beyond (Cambridge
University Press, 1982).
53 Hilary Putnam, The Collapse of the Fact/Value Dichotomy, (Harvard University Press, 2002).
54 Mongin, ‘Value Judgments’.
argument just mentioned. Another response points to a major objection to CBA which
has not been mentioned so far – incommensurability, the view that expressing all
climate change impacts in monetary terms is at best unhelpful, usually misleading,
and in some cases impossible. Taking incommensurability seriously has powerful
implications. Returning to an earlier example, the incommensurability judgement
there – that the impacts of a climate change catastrophe are incommensurably worse
than the consumption losses involved in taking action now – arguably did more work
than the precautionary principle in justifying immediate aggressive action. 55 For
many non-economists, the incommensurability objection to CBA is overwhelming in
the case of climate change. Stern sometimes appears to agree:
Our preference is to consider the multiple dimensions of the cost of climate change
separately, examining each on its own terms. A toll in terms of lives lost gains little
in eloquence when it is converted into dollars; but it loses something, from an ethical
perspective, by distancing us from the human cost of climate change. 56
Nevertheless, Stern does ‘convert into dollars’, in presenting a full CBA which
expresses all impacts on human health, lives lost, and the environment in monetary
terms. (Admittedly, Stern reports both monetary and non-monetary assessments of the
impacts). A discussion of incommensurability, and Stern’s apparently mixed views on
the issue, is far beyond the scope of this paper. 57 Leaving incommensurability aside,
55 Some readers may baulk at the phrase ‘incommensurably worse’. There are two possible
interpretations. First, it is a shorthand for ‘the alternatives are not directly comparable, but on other
ethical grounds one alternative is worse’. Second, the alternatives are comparable (one is worse), but
not commensurable – there is no cardinal measure for ranking them. For an influential statement of the
comparability/commensurability distinction, see
56 Stern, The Economics of Climate Change, pp. 163.
57 The standard arguments used by economists to reject incommensurability problems in CBA are
analysed in Jonathan Aldred, 'Incommensurability and Monetary Valuation', Land Economics, Vol. 82,
No. 2 (2006), pp. 141-61.
any assessment of the role CBA should play in informing climate change policy will
be crucially incomplete.
However, this paper can be interpreted as an exercise in ground-clearing,
before we begin that assessment in earnest. The nature of climate change CBA has
been clarified, showing how it is more an application of applied moral mathematics
than science. With the ground cleared, we can approach the assessment via questions
which will be familiar to institutional political economists: who benefits from
different climate change policies and why? Relatedly, what interests do the costbenefit analysts themselves have? Apart from an institutional analysis of the policy
formation process – different interest groups, their power relations and the impacts of
climate change upon them – political psychology may offer important insights too.
Are arguments couched in cost-benefit terms more or less persuasive than the overtly
‘moral’ arguments which have in the past been used to encourage population-wide
behavioural change?
More prosaically, some key issues in the Stern Review have been examined in
detail. Although it represents a major improvement on past climate change CBA,
stubborn difficulties remain. CBA claims just to value ‘statistical’ lives, but if enough
people are affected, statistical lives add up to real lives lost. The justification for
giving lives a monetary value is to achieve consistent policy-making, which requires a
uniform monetary value. But a uniform value is hard to defend. Ethical disagreements
lie at the heart of debates about the discount rate but the practice of discounting may
itself obscure these disagreements. In particular, use of a single discount rate across
different people and goods is illegitimate, and discounting disenfranchises future
generations. Declining discount rates over time do not address these problems. Pure
uncertainty is a fundamental challenge for climate change policy-making, and it is
hard to see how any purely quantitative approach can address it. In the meantime, the
lesson of Weitzman’s analysis for contemporary political debate is that instead of
focusing on estimating the average impact of climate change, we must pay much more
attention to the possibility of truly horrific catastrophe.
All these difficulties arguably suggest not that we must find better numbers to
feed into climate change CBA, but that we need a different framework entirely, one
which allows explicit ethical debate. If there are persuasive arguments for
incommensurability, this conclusion would be strongly reinforced. But if we persist
with CBA, whether for reasons of political expediency or otherwise, a first step would
be to release the analysis from the meta-ethical straitjacket of attempting to base all
parameters and monetary valuations on market data.